Audio limiting circuit

Abstract

An audio limiting circuit capable of satisfying frequency dependent limits and time domain constraints, is disclosed. In one illustrative embodiment, an input node receives an unattenuated input signal and a system modeling filter predicts the amount, if any, by which the sound pressure level that would be generated by an acoustic transducer in response to the unattenuated input signal, would exceed one or more predetermined limits. In that embodiment, an energy detector separates the excess predicted sound pressure level into one or more frequency bands and calculates the average acoustic energy associated with each band. A gain logic block determines an attenuation factor based on whether one or more of the predetermined limits has been exceeded and the attenuation factor values are smoothed to minimize abrupt changes to the unattenuated input signal. A delay buffer delays the unattenuated input signal values. Finally, in the embodiment described here, the smoothed attenuation factor values are synchronized with and applied to the delayed input signal values and the resulting attenuated signal is transmitted to an output node and ultimately to one or more acoustic transducers.

1. A method of limiting a signal level in a system having one or more predetermined time and frequency domain dependent limits, comprising:
 predicting, at each of a plurality of frequencies, an excess amount by which a sound pressure level that would be output by an acoustic transducer in response to an unattenuated input signal, would exceed at least one of the one or more predetermined time and frequency domain limits; 
 tracking an instantaneous maximum of the excess predicted sound pressure level; 
 calculating an average acoustic energy associated with the excess predicted sound pressure level at each of said plurality of frequencies; 
 calculating at least one attenuation factor based on the average acoustic energy associated with the excess predicted sound pressure level and the instantaneous maximum of the excess predicted sound pressure level; and 
 applying the at least one attenuation factor to an unattenuated input signal. 
 
   
   
     2. The method of  claim 1 , wherein predicting the excess sound pressure level further comprises;
 receiving an unattenuated input signal; 
 determining a model filter transfer function; 
 applying a filter design technique to the model filter transfer function in order to determine non-quantized numerical filter coefficients which describe a time-domain filter; and 
 applying the unattenuated input signal to a time-domain filter described by the non-quantized numerical filter coefficients. 
 
   
   
     3. The method of  claim 2 , wherein determining the model filter transfer function further comprises: determining the model transfer function from the frequency domain characteristics of an acoustic transducer, a digital-to-analog converter, and one or more predetermined limits. 
   
   
     4. The method of  claim 2  wherein the filter design technique applied to the model filter transfer function is the Yule-Walker method. 
   
   
     5. The method of  claim 2  further comprising applying a second transformation to the non-quantized numerical coefficients and quantizing and reformatting the output to produce second order, quantized numerical coefficients of a fixed word length which are compatible with the input requirements of a digital signal processing chip. 
   
   
     6. The method of  claim 5  wherein quantizing and reformatting the second order, non-quantized numerical coefficients further comprises:
 reformatting the non-quantized numerical coefficients from floating-point numbers to fixed-point numbers; 
 truncating the fixed point numbers to a pre-determined number of bits; and 
 formatting the resulting fixed point numbers to make them compatible with the input requirements of a particular signal processing circuit. 
 
   
   
     7. The method of  claim 1  wherein attenuating the unattenuated input signal further comprises:
 delaying the unattenuated input signal; 
 buffering a plurality of attenuation factors; and 
 synchronizing the delayed, unattenuated input signal with the buffered attenuation factors so as to minimize instantaneous changes to the input signal and reduce undesirable acoustic artifacts that could result from instantaneous changes to the input signal. 
 
   
   
     8. The method of  claim 7  wherein buffering the plurality of attenuation factors further comprises:
 determining a target attenuation factor; 
 comparing the target attenuation factor to a reference attenuation factor; and 
 populating a buffer with a plurality of attenuation factors determined by interpolating between the target attenuation factor and the reference attenuation factor. 
 
   
   
     9. The method of  claim 8  wherein, in reference to a digital signal processing system, the reference attenuation factor used for the current time step is the attenuation factor that was applied to the unattenuated, delayed input signal in the previous time step. 
   
   
     10. The method of  claim 8 , wherein linear interpolation is used to populate the buffer with a plurality of attenuation factors. 
   
   
     11. The method of  claim 8  wherein polynomial curve fit interpolation is used to populate the buffer with a plurality of attenuation factors. 
   
   
     12. The method of  claim 8  wherein exponential curve fit interpolation is used to populate the buffer with a plurality of attenuation factors. 
   
   
     13. The method of  claim 1  wherein determining the average acoustic energy associated with each of a plurality of frequency bands of the excess predicted sound pressure level further comprises approximating, for each frequency band, the time-weighted average of the excess acoustic energy associated with each frequency band. 
   
   
     14. The method of  claim 1  wherein determining the attenuation factor further comprises:
 determining a first attenuation as a function of the amount by which the average acoustic energy associated with each frequency band exceeds a first predetermined limit, and the amount of time for which the average acoustic energy associated with any frequency band exceeds the predetermined limit; 
 determining a second attenuation factor as a function of the amount by which the average acoustic energy associated with each frequency band exceeds a second predetermined limit; 
 determining a third attenuation factor as a function of the amount by which the instantaneous maximum amplitude of the predicted sound pressure level exceeds a third predetermined limit; and 
 determining a final attenuation factor as a function of the first attenuation factor, the second attenuation factor and the third attenuation factor. 
 
   
   
     15. The method of  claim 1  wherein the sound pressure that is predicted is not normalized based on predetermined limits such that the result is the sound pressure level that would be output by acoustic transducers in response to the unattenuated input signal. 
   
   
     16. The method of  claim 14  wherein the final attenuation factor is selected so as to minimize the amount of attenuation applied to the unattenuated input signal. 
   
   
     17. An apparatus comprising;
 a first circuit that separates an input electrical representation of a predicted sound pressure level into a predicted sound pressure level in each of a plurality of frequency bands; 
 a second circuit communicatively coupled to the first circuit that calculates the average acoustic energy associated with the predicted sound pressure level in each frequency band; and 
 a third circuit coupled to the second circuit which determines an attenuation factor based on the amount by which the average acoustic energy associated with each frequency band exceeds predetermined limits. 
 
   
   
     18. A signal level limiting apparatus comprising:
 a system modeling filter which predicts at each of a plurality of frequencies an excess amount by which a sound pressure level of an acoustic signal, that would be output by an acoustic transducer in response to an unattenuated input signal, would exceed one or more predetermined limits at each of the plurality of frequencies; 
 a peak detector communicatively coupled to the system modeling filter which tracks an instantaneous maximum level from the predicted sound pressure level received from the system modeling filter; 
 an energy detector which calculates an average acoustic energy associated with the excess predicted sound pressure level at each of said plurality of frequencies; 
 a gain logic block communicatively coupled to the energy detector and the peak detector, which calculates an attenuation factor from the average acoustic energy and the instantaneous maximum levels; and 
 a gain block communicatively coupled to the gain logic block which applies the attenuation factor to the unattenuated input signal. 
 
   
   
     19. A signal level limiting apparatus comprising:
 a spectral energy detector which measures signal energy associated with an unattenuated input signal in a plurality of frequency bands; 
 a peak detector which tracks one or more instantaneous maximum amplitudes associated with the signal energy; 
 an average energy detector which calculates an average acoustic energy associated with the signal energy in each frequency band; 
 a gain logic block communicatively coupled to the spectral energy detector and the peak detector, which calculates an attenuation factor associated with the average acoustic energy and the one or more instantaneous maximum amplitudes; and 
 a gain block communicatively coupled to the gain logic block, which applies the attenuation factor to the unattenuated input signal. 
 
   
   
     20. The apparatus of  claim 18 , further comprising, a delay buffer which delays the unattenuated input signal by a predetermined amount of time. 
   
   
     21. The apparatus of  claim 19 , further comprising, a delay buffer, which delays the unattenuated input signal by a predetermined amount of time. 
   
   
     22. The apparatus of  claim 18  wherein the energy detector is a spectral energy detector that measures the signal energy associated with an unattenuated input signal in a plurality of frequency bands. 
   
   
     23. The apparatus of  claim 22  wherein the spectral energy detector further comprises a bandpass filter bank communicatively coupled to the system modeling filter, wherein the bandpass filter bank separates the excess predicted sound pressure level into a plurality of frequency bands. 
   
   
     24. The apparatus of  claim 20 , further comprising a gain smoother communicatively coupled to the gain logic block and synchronized to the delay buffer, wherein the gain smoother reduces undesirable acoustic artifacts associated with abruptly changing the attenuation factor to be applied to the unattenuated input signal. 
   
   
     25. The apparatus of  claim 21 , further comprising a gain smoother communicatively coupled to the gain logic block and synchronized to the delay buffer, wherein the gain smoother reduces undesirable acoustic artifacts associated with abruptly changing the attenuation factor to be applied to the unattenuated input signal. 
   
   
     26. A method of limiting a signal level, comprising:
 measuring signal energy associated with an unattenuated input signal in a plurality of frequency bands; 
 tracking one or more instantaneous maximum amplitudes associated with the signal energy in each frequency band; 
 calculating an average acoustic energy associated with the signal energy; 
 calculating an attenuation factor associated with the average acoustic energy and the one or more instantaneous maximum amplitudes; and 
 applying the attenuation factor to the unattenuated input signal.